1

Bone age assessment by a quantitative sonometer

Marianna Rachmiel, MD 1,2

, Larisa Naugolni, MD 1, Kineret Mazor-Aronovitch, MD

2,3, Amnon

Levin 4, Nira Koren-Morag, PhD

5, Tzvi Bistritzer, MD

1,2. Ze’ev Hochberg, MD PhD

6

Short title: Bone age by ultrasound

1Pediatric Endocrinology Clinic, Assaf Haroffeh Medical Center, Zerifin Israel,

2Sackler School

of Medicine, Tel Aviv University, Tel Aviv, Israel, 3Pediatric Endocrinology Unit, Shebba

Medical Center, Israel, 4SonicBone Medical Company, Rishon Lezion, Israel ,

5 Department of

Epidemiology and Preventive Medicine, Sackler Faculty of Medicine, Tel-Aviv University. 6

The Ruth and Bruce Rappaport Faculty of Medicine, Technion – Israel Institute of Technology,

Haifa. Israel

Corresponding author: Marianna Rachmiel, Pediatric Diabetes Service, Division of pediatrics,

Assaf-Harofeh Medical Center, Zerifin, 70300, Israel .Tl: 972-8-9542007, Fax: 972-8-9779156,

E.mail:rmarianna@gmail.com

Key terms: skeletal maturation, bone age assessment, SonicBone, sonometer

Conflict of Interest: The study was funded by SonicBone (Rishon Lezion, Israel) .

AL is an employee of SonicBone and ZH is an independent consultant to the company.

Acknowledgements:

The study was funded by SonicBone (Rishon Lezion, Israel).

The authors would like to thank the children and families who participated in this study.

Words count: 2961

2

Abstract

Objective: Bone maturation is currently assessed by radiography using subjective readings or

automated comparison of the characteristics of hand and wrist bones. The aim was to evaluate

the concordance and reproducibility of a new portable device (SonicBone's BAUSTM

) that

utilizes quantitative ultrasonographic (QUS) technology measuring the speed of sound (SOS) of

US waves propagating along measured bone, to current methods Greulich and Pyle (GP) and

Tanner-Whitehouse 3 (TW3).

Methods: The study population included 150 participants, 76 males, age range 4-17 attending an

endocrine clinic. X-ray scans were evaluated independently by four observers. Separate readings

for wrist, carpal and phalanx were averaged for a child's BA assessment. Data from 100 subjects

were utilized to assess BA using speed-of-sound (SOS) and attenuation (ATN) parameters based

on the manual BA by GP method, and to establish a conversion equation for BA assessment by

SonicBone. Fifty subjects were assigned to assess inter modality agreement between the

methods.

Results: The coefficient of determination, R2 ,for the conversion equation including gender, SOS

and ATN was 0.80 for manual GP, 0.81 for automated GP and 0.82 for automated TW3 methods

(p<0.001 for all). There were no significant bias between BA assessment by SonicBone and BA

assessment by all three X-ray based methods. The mean differences of BA assessments were

0.228 (P=0.887), 0.229 (P=0.861) and 0.248 (P=0.274), for manual GP, automated GP and

TW3, respectively.

Conclusion: BA assessment by SonicBone is in agreement to the assessment by all three X-ray

based methods: manual GP, automated GP and TW3.

Introduction:

Skeletal maturity assessment, the so-called ‘bone age’ (BA), is frequently used for evaluating

growth and puberty in children and adolescents. It is recommended as part of the routine clinical

care workup of a child with short or tall stature, precocious and delayed puberty, and more [1, 2].

Repeated BA assessments are an important clinical tool utilized also during the follow-up of

such patients, especially when treated by growth and puberty-related interventions [1, 2]. BA is

currently assessed by radiography of the hand, using eye-balling or automated comparison of the

shape and size of the wrist and hand bones to a standard series of representative radiographic

films of hands according to the Radiographic Atlas of Skeletal Development by Greulich and

3

Pyle (GP) [1, 3], or to the scoring method by Tanner and Whitehouse (TW3), currently in its

2001 third edition [1, 4, 5].

To address the disadvantage of repeated irradiation, the need for specialized radiation centers,

heavy equipment and subjective reading [1, 6-8], a new device, SonicBone (SB) (Rishon Lezion,

Israel) was developed. SB utilizes a quantitative ultrasonographic (QUS) technology of

ultrasonic (US) waves, propagating along a measured bone distance [9, 10]. Here, we report the

validity and reproducibility of BA by SB, as compared to radiographic BA by the GP method

and an automated reading by both the GP and TW3 methods.

Patients and Methods:

Study design: This was a cross sectional study. The study subjects (n=150, 76 males) were

recruited consecutively in a pediatric endocrine clinic. All BAs performed by X-ray scans and

QUS were conducted prior to data analysis. The participants were then randomized to an

’Analysis group’ (n=100, 40 males) and a 'Confirmation group’ (n=50, 27 males). Analysis

performed on data obtained from the ’Analysis group’ resulted in a conversion equation for BA

assessment by SonicBone. Using the conversion equation, data obtained from the ‘Confirmation

group’ assessed the level of agreement, disagreement and confirmation ability between QUS and

the manual GP, automated GP, and automated TW3 methods.

The study has been performed in accordance with the ethical standards as laid down in the 1964

Declaration of Helsinki and its later amendments. The study protocol was approved by the

Institutional Review Board and by the Helsinki Committee of the Israeli Ministry of Health, and

registered at www.clinicaltrials.gov (NCT01346618). Written informed consent was obtained

from each legal guardian, and the participant assented for the study.

Study population: Patients ranging between 4-17 years of age were recruited from the Pediatric

Endocrinology Clinic at Assaf Haroffeh Medical Center, Israel. Inclusion criteria included all

patients who performed hand x-ray scan as part of their clinical care. Exclusion criteria included

children with bone diseases or those who within the last year took medications that might change

bone metabolism or mineralization (such as high dose steroids, biphosphonates, high dose

Vitamin D, or calcitriol).

Data collection: Medical records of the clinical visit performed within 3 months of x-ray

evaluation and of BA assessment were reviewed for clinical and demographic information,

including: gender, age, reason for clinical assessment, concurrent illnesses and medication use,

weight, height, BMI SDS, pubertal stage, and final diagnosis stated for purposes of clinical care.

We reviewed all participants for possible immediate side effects of using the sonic bone device.

4

BA assessment by the manual GP method: Hand x-ray scans were reviewed and assessed

independently by four pediatric endocrinologists being blinded to each other’s findings. They

assigned a separate BA to the long bones (radius and ulna), the carpals and metaphalanges and

phalanges (short bones), and the mean of the three readings was defined as the child’s BA, as

previously described [11, 12] . The mean of four observers was defined as the child’s BA by GP.

Automated BA assessment: The images were analyzed using the BoneXpert version 2.1

automated method for BA determination (Visiana, Denmark), which determines BA by both GP

and TW3 [13, 14]

Ultrasonic BA assessment: The SonicBone (SB) device (Rishon Lezion, Israel) is a small (50cm

X 25cm X 25cm), portable, bone sonometer (Figure 1), which measures two parameters: a)

speed of propagation through bone (speed-of-sound, SOS, m/sec) of inaudible high frequency

waves of a short ultrasound pulse; and b) attenuation (ATN; the decay rate) of the sound wave by

the bone as a function of the distance it travels between a transmitter probe and a receiver probe

located at the edges of the measured bone area. [15]. Three sites of the hand were assessed

according to figure 1: wrist (W), measuring SOS and ATN at the distal radius’ and ulna’s

secondary ossification centers of the epiphyses; metacarpals (MC), measuring SOS and ATN at

the distal metacarpal epiphyses; and phalange (P), measuring SOS and ATN along the bent

proximal third phalanx shaft, growth plate and epiphysis. The average of those three sites was

defined as the child’s BA by SB. All ultrasonic examinations were conducted at the Pediatric

Endocrinology Clinic, by trained personnel. The examiners were blinded to the clinical

background and to the BA by GP or TW3. Each subject underwent two SB readings by two

observers. Eight additional repeated readings were performed for 10 subjects, 5 boys and 5 girls,

aged 6-16 to assess reproducibility and precision positioning assessment.

Statistical Analysis: Data was analyzed by a professional statistician with SPSS software version

21.0 (SPSS Inc. Chicago, Il, USA). The estimation of within subject repeatability was calculated

by a one-way analysis of variance (ANOVA) model. The main analysis of the study included the

correlation and hypothesis testing of equality of BA by SB and BA by manual GP, automated GP

and automated TW3. The immediate side effects were monitored as the numbers of incidents, the

nature of inconveniencies and other complaints. Phase-1 analysis, performed on the entire study

population (n=150), using Pearson correlation coefficient, demonstrated, the linear relationship

between BA by GP and measurements of SB, SOS and ATN parameters. Phase-2 analysis of 100

subjects (‘Analysis group’) established a conversion equation for estimating BA by SB out of the

multivariate linear regression coefficients involving gender and the SB parameters. These

equations provided the best R² result . Phase-3 analysis was performed on data of the

‘confirmation Group’ (n=50) using the equations generated at phase-2 .Comparison of the

5

differences between the BA by SB and by X-ray based methods was conducted by the paired t

test and further presented as 95% confidence intervals. Linear regressions were used to assess the

coefficient of determination (R2) for the BA differences by the mean BA, based on the Bland and

Altman analysis. The linear correlation between BA by X-ray and by SB was further

demonstrated by the Pearson correlation coefficient.

Results

The study population included 150 subjects (76 males), age range (4.1-17.4 years), recruited

between June 2011 and March 2012. At the time of investigation they were diagnosed with short

stature and failure to thrive (46%), growth hormone deficiency (9%), precocious or early puberty

(23%) overweight and obesity (8%), normal and healthy (14%). The clinical, demographic and

body composition characteristics of the analysis and confirmation groups were similar (Table 1).

The SB performance analysis showed high reproducibility and repeatability according to

measurements conducted independently in the same environment by two examiners. Upon

performing 10 repeated readings, on 10 subjects, the percent of relative standard deviation

(%RSD) for SOS were smaller than 0.73% for all the children, with a maximum standard

deviation of 13.7%. The %RSD for ATN was less than 3.5% for all the children with a

maximum standard deviation of 1.4%.

The distribution of SOS and ATN measurements according to skeletal area (W, MC, P) in the

study population (n=150) is presented in Figures 2a and 2b respectively. The SOS measurements

ranged from 1604 - 2647 m/sec and the ATN-surrogate distance ranged from 29.5 - 82.7 mm.

In the phase-1 analysis we correlated between BA by manual and automated GP and automated

TW3 against SOS and ATN from SonicBone in the entire study population (n=150). A

significant correlation for both SOS ( r2=0.68, 0.68, 0.69 for manual GP, automated GP and

automated TW3, respectively) and ATN (r2=0.88, 0.88, 0.89 for e manual GP, automated GP and

the automated TW3, respectively) was found.

In phase-2, multiple linear regression analysis was utilized in order to estimate the most

appropriate predictors for BA by SB , by using all acceptable BA methodologies and the

variables gender, SOS and ATN. The linear regression coefficients are presented in Table 2. For

all three methods, both SOS and ATN were significantly strong predictors for BA. SOS was an

important and significant predictor for BA above and beyond the ATN; the coefficient of

determination R2 significantly increased using SOS. As much as 82% of the total variation in BA

is explained by ATN and SOS. Table 2 also shows the un-standardized coefficient b that are used

6

for predicting future outcomes, and the standardized coefficients beta that were used to evaluate

the relative strength of the relationship to BA.

In the phase-3 analysis, only the data from the ‘confirmation group’ was used. We compared the

assessment of BA by SB, as delivered by the device according to the conversion equation

performed in phase-2, to the assessment of BA from the three hand X-ray based methods (table

3). The differences between the BA by SB and by X-ray based methods were all none significant

(P=0.342, P=0.278 and P=0.229 for manual GP, automated GP and TW3, respectively). The

relatively small R2 (0.186, 0.108 and 0.110 for manual GP, automated GP and TW3,

respectively), based on the Bland and Altman analysis, demonstrate almost no relationship

between disagreement measurements and age. Furthermore, the results demonstrate a significant

correlation between BA by SB and the BA by manual GP (r=0.89, P<0.001), automated GP

(r=0.91, P<0.001) and automated TW3 (r=0.91, P<0.001). .

Discussion

In the dialectics of human anthropology and auxology, BA is an expression of the skeletal

maturity of a child. Inferring from bone maturity, the clinician contemplate diagnostic

considerations in various clinical settings as short stature, delayed and early puberty, as well as

evaluate height prediction, and timing for orthodontic procedures and orthopedic surgery. This is

accomplished in a variety of methods, all of which utilize X-ray technology and compare a given

film to various standards, followed by designation of a BA. The problematical utilization of

repeated X-ray evaluations is well-appreciated. Here, we present the applicability of radiation-

free BA assessment by QUS and its inter-modality agreement with currently employed BA

valuation by X-ray-based methods.

This is not the first attempt to apply QUS technology for BA assessment; other attempts failed to

enter clinical practice. Castriota-Scanderbeg et al [16, 17] attempted to assess skeletal maturation

by quantifying the cartilage overlying layer of the femoral head. They demonstrated a decrease

in cartilage thickness with age. Yet, a comparison with the BA by GP showed poor agreement

[17]. Shimura et al, and Khan et al,[18, 19] assessed skeletal maturation by SOS through a single

site at the head of the ulna (similar to W site in the current study), that often differs from other

bones, not testing the sites available by SB device [2, 11, 12].

The SB device provides three independent measurements of the radius and ulna epiphyses, of

metacarpals and of phalanges, similar to clinical practice assessments of hand x-ray scans. While

the sites assessed by SB and by the X-ray methods are not identical, we demonstrate a significant

inter-modality agreement between BA by SB and BA by GP and TW3 methods at each site

separately and by the mean BA [11].

7

The ultrasound technique used by SonicBone is the "through transmission technique". An

ultrasound wave is propagated perpendicularly through a medium containing soft tissue and bone

from transmitter to receiver. Two parameters are used in this method: The primary parameter is

SOS (time of flight) of the US wave over the distance from transmitter to receiver. SOS

correlates strongly with the structure and density of the bone. However, bone is attenuative and

dispersive. The attenuation is seen in the change in amplitude of a travelling wave and in the

values of the reflection coefficients. The dispersion contributes to the distortion of the wave, and

we therefore included in the equation the attenuation – the decay rate of a wave as it propagates

through bone, and use the distance between transmitter and receiver as an ‘Attenuation factor -

ATN’.

The BA by SB was generated by the equation for BA assessment by QUS, which was integrated

in the device, according to the data retrieved from the analysis of 100 subjects, including SOS,

ATN and the manual reading by the GP method. It was then confirmed in 50 subjects against

both manual and automated GP reading as well as the automated TW3 methods showing a high

performance of reliability and significant agreement.

For all three methods, the ATN made an important contribution to the regression, yet, compared

to all three x-ray based the SOS was important and significant predictor for bone age above and

beyond the ATN.

Thus, the measurements by SB are all hand-area inclusive (W, MC, P), objective, and of

physiological agreement to the goal of bone maturation assessment, offering a possible

alternative to the present radiation based mostly subjective GP, and TW3 methods. Beyond the

carpals’ primary- and the metacarpals’ secondary ossification centers, BA by SB includes the

primary ossification centers of the phalanx. As BA is an essential measurement procedure for

pediatric endocrinology physicians and is often repeated over time, SB offers an important

advantage over the current methods, radiation-free and with objective readings by a device

accessible at the clinician office. Furthermore, no side effects or dissatisfaction were dettected

among participants aftre using the SB equipment.

A limitation of our study is the lack of reference or standard for BA by chronological age for the

QUS method. This is currently under development using a normal population distribution for all

ages according to gender.

In summary, the radiation-free assessment of BA by SB device in a wide age pediatric

population attending endocrine clinics, using assessment of wrist, carpal and phalangeal areas,

was found to be highly reproducible, comparable and in agreement to BA assessed by all

conventional X-rays based methods.

8

Ethical approval:

All procedures performed in studies involving human participants were in accordance with the

ethical standards of the institutional and/or national research committee and with the 1964

Helsinki declaration and its later amendments or comparable ethical standards.

References

1. Martin DD, Wit JM, Hochberg Z, et al. (2011) The use of bone age in clinical practice -

part 1. Horm Res Paediatr 76:1-9.

2. Martin DD, Wit JM, Hochberg Z, et al. (2011) The use of bone age in clinical practice -

part 2. Horm Res Paediatr 76:10-16.

3. Greulich W PS (1959) Radiographic atlas of skeletal development of the hand and wrist.

Stanford university press.

4. Tanner JM WR (1937) The atlas of skeletal maturation. Mosby Company, St. Louis, .

5. Tanner JM WR, Cameron N, Marshall WA, Healy MJ, Goldstein H. (ed) (1975)

Assessment of skeletal maturity and prediction of adult height: TW2 method. Academic

Press, New York.

6. Bull RK, Edwards PD, Kemp PM, et al. (1999) Bone age assessment: a large scale

comparison of the Greulich and Pyle, and Tanner and Whitehouse (TW2) methods. Arch

Dis Child 81:172-173.

7. Johnson GF, Dorst JP, Kuhn JP, et al. (1973) Reliability of skeletal age assessments. Am

J Roentgenol Radium Ther Nucl Med 118:320-327.

8. Lynnerup N, Belard E, Buch-Olsen K, et al. (2008) Intra- and interobserver error of the

Greulich-Pyle method as used on a Danish forensic sample. Forensic Sci Int 179:242

e241-246.

9. Specker BL, Schoenau E (2005) Quantitative bone analysis in children: current methods

and recommendations. J Pediatr 146:726-731.

10. Zadik Z, Price D, Diamond G (2003) Pediatric reference curves for multi-site quantitative

ultrasound and its modulators. Osteoporos Int 14:857-862.

11. Hochberg Z (2002) Endocrine control of skeletal maturation: Annotation to bone age

readings. Karger Medical and Scientific Publishers.

12. Even L, Andersson B, Kristrom B, et al. (2014) Role of growth hormone in

enchondroplasia and chondral osteogenesis: evaluation by X-ray of the hand. Pediatr Res

76:109-114.

13. Thodberg HH, Kreiborg S, Juul A, et al. (2009) The BoneXpert method for automated

determination of skeletal maturity. IEEE Trans Med Imaging 28:52-66.

9

14. Martin DD, Deusch D, Schweizer R, et al. (2009) Clinical application of automated

Greulich-Pyle bone age determination in children with short stature. Pediatr Radiol

39:598-607.

15. Hosokawa A, Otani T (1997) Ultrasonic wave propagation in bovine cancellous bone.

The Journal of the Acoustical Society of America 101:558-562.

16. Castriota-Scanderbeg A, De Micheli V (1995) Ultrasound of femoral head cartilage: a

new method of assessing bone age. Skeletal Radiol 24:197-200.

17. Castriota-Scanderbeg A, Sacco MC, Emberti-Gialloreti L, et al. (1998) Skeletal age

assessment in children and young adults: comparison between a newly developed

sonographic method and conventional methods. Skeletal Radiol 27:271-277.

18. Khan KM, Miller BS, Hoggard E, et al. (2009) Application of ultrasound for bone age

estimation in clinical practice. J Pediatr 154:243-247.

19. Shimura N KS, Arisaka O, Imataka M, Sato K, Matsuura M (2005) Assessment of

measurement of children's bone age ultrasonically with Sunlight BonAge. Clin Pediatr

Endocrinol 14 (Suppl 24):17-20.

Figure legends:

Figure 1: SonicBone device . An illustration of the hand locations measured according to hand

radiograph for comparison, and an illustration of device parameters. Measurements are

performed at third proximal phalange, the metacarpal and the wrist. The device measures the

speed of propagation of inaudible high frequency waves of a short ultrasound pulse through

bone, and the Attenuation/distance between a transmitter probe (T) and a receiver probe (R),

located at the edges of the measured bone area for assessment of the attenuation factor.

Figure 2a: Quartiles distribution of SonicBone parameters of SOS (speed of sound) in m/sec,

according to the measured areas of the left hand, wrist (W), carpal (MC) and phalanx (P), in the

whole study population (n=150). Lines within boxes indicate median; limits of boxes indicate

25th and 75th percentiles; circles represent outliers.

Figure 2b: Quartiles distribution of SonicBone parameters of attenuation/distance in mm ATN,

according to the measured areas of the left hand, wrist (W), carpal (MC) and phalanx (P), in the

whole study population (n=150). Lines within boxes indicate median; limits of boxes indicate

25th and 75th percentiles; circles represent outliers.

10

Table 1: Demographic, clinical and body composition parameters of study population, randomly

divided into the analysis and the confirmation groups

All ‘Analysis

Group’

‘Confirmation

Group’

P

value

Number 150 100 50

Gender (f) 74 51 (51%) 23 (46%) 0.34

Pre-puberty* 46 39 (39%) 17 (34%) 0.33

BMI SDS 0.21.4 0.31.3 -0.11.5 0.09

Age (years) 10.63.3 10.53.2 10.93.4 0. 85

Mean BA by GP - W (years) 10.03.4 10.13.2 10.03.8 0.20

Mean BA by GP - CMC

(years)

10.13.5 10.13.3 10.03.8 0.26

Mean BA by GP - P (years) 10.33.4 10.33.2 10.33.7 0.47

Mean BA by GP (all sites ) 10.1 3.3 10.1 3.1 9.9 3.6 0.77

*Data is based on n=148, FTT- failure to thrive, GHD – growth hormone deficiency.

Table 2: The regression coefficients, used in the conversion equation from the Analaysis Group

Difference between

BA methods

Parameter

Coef.b SE Beta t P value R2

SB-Manual GP

Constant -28.68 4.87 -5.88 <0.001

0.80

Gender 0.187 0.310 0.030 0.602 0.548

SOS *(cm/sec) 1.09 0.293 0.230 3.73 <0.001

ATN **(mm) 0.330 0.029 0.724 11.39 <0.001

SB-Automated GP

Constant -30.77 4.88 -6.30 <0.001

0.81

Gender 0.345 0.31 0.053 1.11 0.268

SOS *(cm/sec) 1.160 0.293 0.236 3.96 <0.001

ATN **(mm) 0.341 0.029 0.722 11.75 <0.001

SB- Automated TW3

Constant -29.55 4.55 -6.49 <0.001

0.82

Gender 0.411 0.289 0.067 1.42 0.159

SOS *(cm/sec) 1.133 0.274 0.241 4.14 <0.001

ATN **(mm) 0.323 0.027 0.718 11.94 <0.001

11

BA- bone age, SB - BA by SB, GP - Greulich and Pyle, TW3= Tanner and Whitehouse , SE=

standard error, t=statistics value. *SOS- speed of sound, measurements ranges from 1,604 m/sec

to 2,647 m/sec. ***ATN- attenuation, measurements ranges from 29.5 mm to 82.7 mm.

Table3: Comparison of BA by SonicBone to the BA by the three X-ray methods (in the

confirmation group of subjects, n=50).

Mean

BA

Mean

difference

SD

S.E.

95% C.I.

t

P

R2

SB - manual GP

9.84

0.228

1.68

0.24

-0.25-0.71

-0.960

0.342

0.186

SB - automated GP

9.77

0.229

1.47

0.21

-0.19-0.65

01.097

0.278

0.108

SB-automated TW3

9.57

0.248

1.44

0.20

-0.16-0.66

1.218

0.229

0.110

Figure 1

12

Figure 2a

Figure 2b